A Multilevel Modeling Approach Towards Wind Farm Aggregated Power Curve

Wind farm multiple aggregated power curve modeling plays an important role in reducing the complexity of analyses in wind farm management and annual power prediction. There is a trade-off between the complexity and accuracy of aggregated power curves. In this paper, $K$-Means clustering is utilized to classify turbines in a wind farm into homogeneous groups according to a new set of features based on the overall performance of turbines. We apply multilevel modeling methods, including random intercept and random slope models on turbine clusters, to take into account the hidden correlation among different clusters. Results show that the accuracy of our proposed methods are higher than the single aggregated method alongside an equal complexity. The proposed multiple aggregated power curve model can be utilized to analyze wind farm behavior and wind farm power simulations to forecast wind power.